← back to index

g00: contempt disdain loathing and detestation

Study: best-of-100 voice-acting selection across speaker identities. Model: an 8B autoregressive expressive TTS model, generated WITH reference-voice conditioning (audio temperature 1.0, repetition penalty 1.1, text temperature 0.7, top-p 0.95, top-k 25, bf16). Reference source: a large open crowd-sourced corpus of expressive human speech (one distinct speaker clip per group, diverse emotions). Reward: combined = invWER x (norm_genu + norm_blend), where genu is a commercial-voice genuineness score (0-6) and blend an expressive-quality score (0-10), each min-max normalized across all generated clips; invWER = max(0, 1-WER) from multilingual ASR. Achieved throughput: 2.87 clips/s on a single GPU.

English script: You actually think your little apology means anything to me now? I have met more honest people crawling in the gutter than I ever found in your company. Do not ever mistake my silence for respect, because you have earned none of it.

German script (translation): Du glaubst tatsaechlich, deine kleine Entschuldigung wuerde mir jetzt noch irgendetwas bedeuten? Ich habe ehrlichere Menschen in der Gosse getroffen, als ich je in deiner Gesellschaft gefunden habe. Verwechsle mein Schweigen niemals mit Respekt, denn den hast du dir in keiner Weise verdient.

Delivery instruction: Cold contempt and disdain, the voice low and sneering, dripping with icy loathing. High quality recording.

Original reference voice

REFgenu=0.53 blend=0.44

English — top 8 / bottom 2 of 100

#1combined=0.558 blend=1.46 genu=1.42 invWER=1.00
#2combined=0.446 blend=1.21 genu=1.60 invWER=0.79
#3combined=0.426 blend=1.01 genu=1.16 invWER=0.98
#4combined=0.392 blend=0.74 genu=1.13 invWER=1.00
#5combined=0.378 blend=0.95 genu=0.94 invWER=1.00
#6combined=0.374 blend=0.87 genu=1.01 invWER=0.98
#7combined=0.372 blend=0.00 genu=1.50 invWER=1.00
#8combined=0.371 blend=0.80 genu=1.00 invWER=1.00
#99 (bottom)combined=0.032 blend=2.33 genu=1.44 invWER=0.05
#100 (bottom)combined=0.027 blend=1.72 genu=1.35 invWER=0.05

German — top 8 / bottom 2 of 100

#1combined=0.373 blend=0.47 genu=1.30 invWER=0.95
#2combined=0.345 blend=1.12 genu=0.77 invWER=0.95
#3combined=0.324 blend=0.50 genu=1.12 invWER=0.91
#4combined=0.302 blend=0.51 genu=1.05 invWER=0.88
#5combined=0.301 blend=0.11 genu=1.18 invWER=0.95
#6combined=0.281 blend=0.93 genu=0.62 invWER=0.93
#7combined=0.270 blend=0.23 genu=0.96 invWER=0.95
#8combined=0.258 blend=0.27 genu=0.92 invWER=0.93
#99 (bottom)combined=0.042 blend=2.71 genu=2.19 invWER=0.05
#100 (bottom)combined=0.000 blend=6.21 genu=1.88 invWER=0.00

Voice-converted to a random speaker A — source sped +2%, target=Source-A_de_04

REFgenu=1.37 blend=1.18

English — top 8 / bottom 2 of 100

#1combined=0.759 blend=1.20 genu=2.46 invWER=1.00
#2combined=0.733 blend=1.90 genu=2.08 invWER=0.95
#3combined=0.611 blend=1.30 genu=1.76 invWER=1.00
#4combined=0.580 blend=1.09 genu=1.75 invWER=1.00
#5combined=0.569 blend=0.69 genu=1.94 invWER=1.00
#6combined=0.565 blend=0.27 genu=2.18 invWER=1.00
#7combined=0.556 blend=0.89 genu=1.76 invWER=1.00
#8combined=0.555 blend=0.00 genu=2.30 invWER=1.00
#99 (bottom)combined=0.046 blend=0.00 genu=0.06 invWER=1.00
#100 (bottom)combined=0.031 blend=0.00 genu=-0.00 invWER=1.00

German — top 8 / bottom 2 of 100

#1combined=1.907 blend=7.20 genu=4.25 invWER=0.95
#2combined=1.406 blend=5.25 genu=3.30 invWER=0.93
#3combined=1.396 blend=5.13 genu=3.86 invWER=0.86
#4combined=1.322 blend=4.35 genu=3.45 invWER=0.93
#5combined=1.315 blend=4.82 genu=3.29 invWER=0.91
#6combined=1.286 blend=4.42 genu=3.09 invWER=0.95
#7combined=1.195 blend=4.27 genu=3.36 invWER=0.86
#8combined=1.179 blend=3.94 genu=3.17 invWER=0.91
#99 (bottom)combined=0.000 blend=0.00 genu=-0.14 invWER=0.95
#100 (bottom)combined=0.000 blend=4.51 genu=2.36 invWER=0.00

Voice-converted to a random speaker B — source sped +47%, target=Source-B

REFgenu=1.59 blend=0.00

English — top 8 / bottom 2 of 100

#1combined=0.739 blend=0.30 genu=2.92 invWER=1.00
#2combined=0.707 blend=0.37 genu=2.74 invWER=1.00
#3combined=0.670 blend=0.75 genu=2.35 invWER=1.00
#4combined=0.646 blend=0.00 genu=2.70 invWER=1.00
#5combined=0.644 blend=1.17 genu=1.97 invWER=1.00
#6combined=0.629 blend=1.22 genu=1.88 invWER=1.00
#7combined=0.606 blend=0.25 genu=2.37 invWER=1.00
#8combined=0.604 blend=0.91 genu=1.95 invWER=1.00
#99 (bottom)combined=0.209 blend=0.27 genu=0.93 invWER=0.74
#100 (bottom)combined=0.151 blend=0.00 genu=1.10 invWER=0.53

German — top 8 / bottom 2 of 100

#1combined=0.883 blend=2.44 genu=2.54 invWER=0.93
#2combined=0.811 blend=2.39 genu=2.14 invWER=0.95
#3combined=0.790 blend=2.02 genu=2.27 invWER=0.95
#4combined=0.648 blend=1.38 genu=2.01 invWER=0.95
#5combined=0.646 blend=1.29 genu=2.05 invWER=0.95
#6combined=0.625 blend=1.31 genu=1.94 invWER=0.95
#7combined=0.553 blend=0.84 genu=1.96 invWER=0.93
#8combined=0.514 blend=0.82 genu=1.73 invWER=0.95
#99 (bottom)combined=0.055 blend=0.00 genu=0.12 invWER=0.93
#100 (bottom)combined=0.000 blend=3.58 genu=1.52 invWER=0.00